Genetically regulated epigenetic transcriptional activation of insertion confers mouse dactylaplasia phenotype

Hiroki Kano*, Hiroki Kurahashi†, and Tatsushi Toda*‡

*Division of Clinical Genetics, Department of Medical Genetics, Osaka University Graduate School of Medicine, Suita, Osaka 565-0871, Japan; and †Division of Molecular Genetics, Institute for Comprehensive Medical Science, Fujita Health University, Toyoake 470-1192, Japan

Edited by Mark T. Groudine, Fred Hutchinson Cancer Research Center, Seattle, WA, and approved September 7, 2007 (received for review June 12, 2007)

Dactylaplasia, characterized by missing central digital rays, is an were identified at 10q24 in several SHFM3 families (10–13). The inherited mouse limb malformation that depends on two genetic smallest duplicated region contained a disrupted extra copy of loci. The first locus, Dac, is an insertional mutation around the the dactylin gene and the LBX1, BTRC, POLL, and DPCD genes dactylin gene that is inherited as a semidominant trait. The second in their entirety. The dactylin gene encodes an F-box/WD40 locus is an unlinked modifier, mdac/Mdac, that is polymorphic repeat protein; members of this protein family commonly func- among inbred strains. Mdac dominantly suppresses the dactylapla- tion in ubiquitin-dependent proteolytic pathways (14). Although sia phenotype in mice carrying Dac. However, little is known about the dactylin gene is considered to be the best candidate for either locus or the nature of their interaction. Here we show that SHFM3 and the mouse dactylaplasia phenotype, its specific Dac is a LTR retrotransposon insertion caused by the type D mouse function remains undetermined. In mice, the Dac2J insertion in endogenous provirus element (MusD). This insertion exhibits dif- intron 5 of dactylin results in the absence of the normal dactylin ferent epigenetic states and spatiotemporally expresses depend- mRNA transcript; in contrast, the Dac1J insertion in the up- ing on the mdac/Mdac modifier background. In dactylaplasia mu- stream region affects neither the size nor the amount of dactylin -tants (Dac/؉ mdac/mdac), the LTRs of the insertion contained transcript (3). Furthermore, human SHFM3 patients demon unmethylated CpGs and active . Furthermore, MusD strate only partial dactylin duplications (10, 11). Therefore, the elements expressed ectopically at the apical ectodermal ridge of possibility exists that the dactylin gene itself is simply a bystander limb buds, accompanying the dactylaplasia phenotype. On the and that Dac insertions have long-range regulatory effects on .other hand, in Dac mutants carrying Mdac (Dac/؉ Mdac/mdac), the neighboring genes 5؅ LTR of the insertion was heavily methylated and enriched with The mouse dactylaplasia phenotype depends not only on the inactive chromatin, correlating with inhibition of the dactylaplasia genotype at the mutated Dac locus but also on homozygosity for phenotype. Ectopic expression was not observed in the presence of a recessive allele in another unlinked locus, mdac, which has Mdac, which we refined to a 9.4-Mb region on mouse chromosome been mapped to chromosome 13, between D13Mit10 and 13. We report a pathogenic mutation caused by MusD. Our findings D13Mit99 (2). This locus is polymorphic among inbred strains, indicate that ectopic expression from the MusD insertion correlates and two alleles have been identified. Inbred strains such as with the dactylaplasia phenotype and that Mdac acts as a defen- BALB/cJ, A/J, and 129/J carry mdac, which permits Dac expres- sive factor to protect the host from pathogenic MusD sion; on the other hand, inbred strains such as CBA/J, C3H/J, and insertions. C57BL/6J carry the Mdac allele, which dominantly inactivates Dac (2). The Dac1J and Dac2J alleles are equally sensitive to Mdac dactylin ͉ ectrodactyly ͉ LTR ͉ split hand/split foot malformation ͉ (3). Therefore, the dactylaplasia phenotype is observed in only methylation mice homozygous for mdac (Dac/ϩ mdac/mdac or Dac/Dac mdac/mdac) and is never observed in mice carrying Mdac, actylaplasia is an inherited mouse limb malformation that is regardless of their Dac status (Dac/ϩ Mdac/mdac or Dac/Dac Dcharacterized by missing central digital rays. The Dac mutation Mdac/mdac). Mutational insertions in the Dac locus have been is inherited as a semidominant trait, evidenced by missing central identified and partially cloned; however, little is known about the digits in the fore- and hindlimbs of heterozygous mice and mono- insertional mutation or its modifier. dactyly in homozygous mice (1, 2). The Dac locus has been mapped The present study aimed to characterize Dac, Mdac, and to the distal end of chromosome 19. Two independent Dac mutant interactions between the two loci to elucidate the pathological alleles, Dac1J and Dac2J, arose spontaneously in breeding colonies. mechanism of dactylaplasia. Our study found that both Dac Both are insertions residing within the same locus: Dac1J is located in the region upstream of the dactylin gene, and Dac2J is located in intron 5 of dactylin (3). Southern blot analysis indicated that both Author contributions: H. Kano, H. Kurahashi, and T.T. designed research; H. Kano per- formed research; H. Kano analyzed data; and H. Kano and T.T. wrote the paper. insertions are larger than 4.5 kb; however, ‘‘jumping PCR’’ identi- fied only the LTR portion of the insertion (3). Partial PCR products The authors declare no conflict of interest. terminating in the 5Ј and 3Ј LTRs cross-prime each other, resulting This article is a PNAS Direct Submission. in amplified products that lack any of the internal sequence between Abbreviations: AER, apical ectodermal ridge; E, embryonic day; ETn, early transposon; MusD, the type D mouse endogenous provirus element; SHFM, split hand/split foot LTRs. Therefore, these insertions were thought to be caused by an malformation. early transposon (ETn) element, which is a common mutagen in Data deposition: The sequences reported in this paper have been deposited in the DNA mice (4, 5). Data Bank of Japan (accession nos. AB305072 and AB305073). Split hand/split foot malformation (SHFM) in humans is a See Commentary on page 18879. congenital limb malformation that has an ectrodactyly pheno- ‡To whom correspondence should be addressed at: Division of Clinical Genetics, Depart- type analogous to that of the dactylaplasia mouse. SHFM is ment of Medical Genetics, Osaka University Graduate School of Medicine, 2-2-B9, Yamad- genetically heterogeneous; to date, five different loci, SHFM1 to aoka, Suita, Osaka 565-0871, Japan. E-mail: [email protected]. SHFM5, have been mapped. Dactylaplasia is a mouse model of This article contains supporting information online at www.pnas.org/cgi/content/full/ SHFM3 because the Dac locus is syntenic to the SHFM3 locus 0705483104/DC1. (10q24) (2, 6–9). Recently, 0.5-Mb tandem genomic duplications © 2007 by The National Academy of Sciences of the USA

19034–19039 ͉ PNAS ͉ November 27, 2007 ͉ vol. 104 ͉ no. 48 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705483104 Downloaded by guest on September 30, 2021 A 8fgF nilytcad

exon 1 2 3 4 5 6 7 8 9 SEE COMMENTARY leT Cen

aD c J1 i itresn on 5‘ RTL caD J2 noitresni 3‘ RTL TG TAT G G TATT G AAAGAG AAAGAG 3‘ RTL 5‘ RTL exon 1 2 3 4 5 6 7 8 9 eT l Cen

B caD 1J sucol caD 2J col us C orp 5‘ LTR gag op l 3‘ RTL 2J 1J Dac Dac / / 2J 1J O O 2 2 M2 H M1 +/+ +/+ H Dac Dac

)pb( )pb(

031,32

,9 4 61 (AF246633) 6 755, 63,4 1 MusD

0,1 00 058 6 05 005 D ca 1J sni e noitr

Fig. 1. Characterization of Dac insertions. (A) Insertions of LTR around the dactylin gene. The Dac1J insertion was integrated 10 kb upstream of the dactylin gene in antisense orientation. The Dac2J insertion was integrated into intron 5 of the dactylin gene in sense orientation. (B) PCR amplification of the insertions. The Dac1J and Dac2J insertions were 7,486 and 7,473 bp, respectively, and were 99.6% identical in sequence. M1, Lambda DNA-HindIII digest (New England Biolabs, Beverly, MA); M2, 1Kb Plus DNA Ladder (Invitrogen). (C) Dot plot DNA comparison of the Dac1J insertion and the MusD element (AF246633). The stringency of comparison was 19 of 23. Each Dac insertion shared 100% identity within the 5Ј and 3Ј LTRs and contained intact ORFs for the gag, pro, and pol genes.

insertions are caused by a type D mouse endogenous provirus their identical genotypes, suggesting an epigenetic effect on (MusD) element. We observed a correlation between the dac- phenotype severity. To investigate the epigenetic status of these tylaplasia phenotype and the epigenetic status of the MusD insertions, bisulfite sequencing and ChIP studies were per- insertion, which is modulated by its modifier. Furthermore, we formed for the Dac1J insertion by using freshly prepared em- show ectopic expression of MusD and its related elements in Dac bryonic tissues. An unrelated MusD element (AL773522) in the mutant limb buds. These observations demonstrate the impact of mouse genome was used as a control for each experiment. This retrotransposon insertions in the genome, as well as a host element shows the greatest sequence similarity to Dac insertions defensive mechanism against retrotransposons. in the mouse genome; its 5Ј and 3Ј LTRs are 100% identical, containing 18 CpGs, but it has an ORF-disrupting mutation in Results the pol gene due to a 1-bp deletion. It is known that most LTR Characterization of Dac Insertions. Two independent, spontane- 1J 2J retrotransposons in somatic cells are maintained in a heavily ously arising Dac mutant alleles, Dac and Dac , are caused by methylated and silent state (16). As expected, bisulfite sequenc- insertions around the dactylin gene. Each was previously par- ing showed that both the 5Ј and 3Ј LTRs of the AL773522 tially cloned (3). To further characterize these insertions, we element were heavily methylated in Dac mutant mice (Dac1J/ϩ GENETICS isolated each by PCR and sequenced them directly (Fig. 1A). Our mdac/mdac) (Fig. 2A, Top). However, the Dac1J insertion carried analysis identified the insertions as LTR retrotransposons with an unmethylated 5Ј LTR and relatively hypomethylated 3Ј LTR lengths of 7,486 bp (Dac1J insertion; AB305072) and 7,473 bp (Dac2J insertion; AB305073), each containing 6-bp target-site even in somatic cells, exhibiting some interclone and intermouse duplications (Fig. 1B). Interestingly, the two sequences were variation (Fig. 2A, Middle). We saw no significant difference in 99.6% identical, and they shared sequence identity with an LTR DNA methylation status of the several tissue types analyzed for retrotransposon (AF246633) that was originally reported as a this study, including brain, liver, kidney, and tail (data not MusD element (15) (Fig. 1C). The Dac1J insertion was integrated shown). 10 kb upstream of the dactylin gene in antisense orientation, The ChIP assay revealed a correlation between DNA meth- whereas the Dac2J insertion was integrated in intron 5 of dactylin ylation and modification status within the LTRs of the in sense orientation (Fig. 1A). The Dac2J insertion site was Dac1J insertion. The unmethylated 5Ј LTR of the Dac1J insertion identified at position 121,540 bp on AC003694 by inverse PCR (Dac1J/ϩ mdac/mdac) consisted of active chromatin enriched before isolation of the Dac2J insertion site (data not shown). with acetylated histone and H3-Lys-4 methylation (Fig. 2B). We Each insertion had 100% identical 5Ј and 3Ј LTRs and contained also performed bisulfite sequencing analysis on primary cultured intact ORFs for gag, pro, and pol genes. embryonic fibroblasts but found that the DNA methylation status in fibroblasts differed from that of original fresh tissues, DNA Methylation and Histone Modification of Dac Insertions. Dac- even after several passages (data not shown). Modified epige- tylaplasia mice show a wide range of phenotypic variation despite netic status sometimes is observed during the establishment of

Kano et al. PNAS ͉ November 27, 2007 ͉ vol. 104 ͉ no. 48 ͉ 19035 Downloaded by guest on September 30, 2021 A cadm / cadm cadM / cadm +/+ caD J1 /+ caD J1 / caD J1 caD 1J/ caD 1J

AL773522

DsuM

5‘ LTR 3‘ LTR

gF f8 (mdac/mdac)

il bm Dac1J insertion

A P

Fig. 3. Abnormal expression of MusD/ETn elements in dactylaplasia mutant (Mdac/mdac) limb buds. In situ hybridization to assess MusD/ETn and Fgf8 expression in whole embryos and their anterior limb buds collected at E10.5 and the representative phenotype of each genotype. Arrows indicate aberrant ex- pression in the AER of the dactylaplasia mutant limb bud.

5‘ LTR 3‘ LTR B this ectopic expression was more apparent and more widely tnemhcirne evitaleR tnemhcirne 6 30 6 30 dispersed across the AER. Expression was strongest in anterior 5 5 limb buds on E10.5 and in posterior limb buds on E11 because 4 20 4 20 3 3 there is a 0.5-day developmental delay of the hindlimbs with 2 10 2 10 1 1 respect to the forelimbs (data not shown). These mutants showed 0 0 0 0 no aberrant expression other than in limb buds. H4Ac H3Ac H3mK4 H3mK9 H4Ac H3Ac H3mK4 H3mK9 Effect of Mdac. To investigate the effect of Mdac, dactylaplasia Fig. 2. Epigenetic modification of the Dac insertion. (A) DNA methylation mice were crossed with C57BL/6J mice that carry Mdac. DNA 1J profile for each CpG in the LTRs of the Dac insertion. Bisulfite-treated from F1 hybrid embryos (Dac1J/ϩ Mdac/mdac) was subjected to genomic DNA derived from Dac mutants in either the mdac/mdac or Mdac/ Ј mdac background was subjected to LTR amplification. PCR fragments were bisulfite sequencing. Contrary to the unmethylated 5 LTR seen 1J ϩ Ј subcloned and sequenced. LTRs of the MusD element on AL773522 were in the dactylaplasia mutant (Dac / mdac/mdac), the 5 LTR of 1J analyzed as a control. Open circles, unmethylated CpGs; filled circles, meth- the Dac insertion in the Mdac background was heavily meth- ylated CpGs. Each line represents the sequence of one clone, and each block ylated, although the 3Ј LTR was unaffected (Fig. 2A, Bottom). of clones represents the data from one mouse. (B) ChIP analysis of LTRs from We saw no significant difference in methylation status between the Dac1J insertion in either the mdac/mdac (white columns) or Mdac/mdac the sense and antisense strands of either LTR (data not shown). (black columns) background. Mononucleosomes were prepared from fresh The methylated 5Ј LTR of the Dac1J insertion (Dac1J/ϩ Mdac/ embryos and subjected to ChIP by using antibodies against acetylated H4, mdac) was depleted of histone acetylation but was enriched for acetylated H3, methylated H3-Lys-4 (H3mK4), and methylated H3-Lys-9 Ј (H3mK9). Quantitative PCR analysis with SYBR Green was performed, and the methylated H3-Lys-9 (Fig. 2B). The 3 LTR showed no change data were summarized after normalizing either to the MusD element on in histone modification status in the presence of Mdac, just as its AL773522 [acetylated histone H4 (H4Ac), acetylated histone H3 (H3Ac), and DNA methylation status was unaffected. H3mK4] or Actb (H3mK9); levels for both were set to 1.0. Ectopic expression seen in the dactylaplasia mutant (Dac1J/ Dac1J mdac/mdac) was not observed in embryos carrying Mdac (Dac1J/Dac1J Mdac/mdac), correlating with its lack of a dacty- a cell line, which does not represent the original status (17). laplasia phenotype (Fig. 3). Fgf8 was used as a control AER Therefore, only embryonic tissue was used for both bisulfite marker because dactylaplasia is known to demonstrate partial sequencing and the ChIP assay. loss of Fgf8 expression in the AER of mutant limb buds (3, 19). Fgf8 expression was restored in embryos with Mdac, which Ectopic Expression of MusD Elements in Dactylaplasia Mutant Mice. exhibited no ectopic expression of MusD. Recently, a tissue- and stage-specific expression pattern for The Dac2J mutant shared features with the Dac1J mutant. The MusD and ETn elements has been shown in organogenesis-stage 5Ј LTR of the Dac2J insertion also showed two contrasting mouse embryos (18). To examine the MusD expression pattern epigenetic states dependent on the mdac/Mdac modifier back- in dactylaplasia mice, we generated a MusD antisense strand ground. In situ hybridization of Dac2J mutant embryos similarly probe from the 3Ј UTR region of the MusD element for showed aberrant MusD expression and loss of Fgf8 expression in hybridization to embryos in various developmental stages [em- the AER, which were restored by the presence of Mdac (data not bryonic day (E)9.5–E11.5] (Fig. 3). Expression was observed in shown). both fore- and hindlimb buds and confined to the mesenchyme The Dac2J mutation is characterized by the absence of the in WT embryos, as reported previously (18). Interestingly, normal dactylin mRNA transcript (3). To see whether the heterozygous dactylaplasia mutant (Dac1J/ϩ mdac/mdac) em- dactylin transcript was restored in the presence of the Mdac, bryos exhibited aberrant MusD expression at the central portion RNA derived from E10.5 embryos was subjected to quantitative of the apical ectodermal ridge (AER) in addition to its normal real-time RT-PCR. To our surprise, Mdac did not affect the expression. In homozygous mutants (Dac1J/Dac1J mdac/mdac), dactylin gene. The dactylin transcript in the Dac2J mutant was

19036 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705483104 Kano et al. Downloaded by guest on September 30, 2021 922A1 909A9 A 922A1 922A3 922A6 B (Dac/+) (Dac/Dac) SEE COMMENTARY Phenotype NN D

Genotype Dac2J/++/+ Dac2J/+ (Dac locus)

D13Mit20

m/m M/mm/m locus

D13Mit113 Mdac

m/m M/mm/m Inferred

D13Mit209

M/m M/mm/m

Fig. 4. Mapping of Mdac on mouse chromosome 13. F1 hybrids (Dac/ϩ Mdac/mdac) were produced by crossing the dactylaplasia mouse (Dac/ϩ mdac/mdac) and C57BL/6J (ϩ/ϩ Mdac/Mdac) and then backcrossed to the dactylaplasia parental strain. These backcrosses were typed for both Dac and microsatellite markers on chromosome 13, and phenotypes were compared. (A) Recombination between D13Mit113 and D13Mit209 in a mouse 922A1. The mouse 922A1 exhibited normal phenotype despite carrying the Dac2J mutant allele. Microsatellite markers that were homozygous (D13Mit20 and D13Mit113) were excluded from the Mdac locus because this mouse must have the Mdac allele. N, normal; D, dactylaplasia; m, allele inherited from the dactylaplasia parental strain; and M, allele inherited from the C57BL/6J strain. (B) Recombination events in two offspring (922A1 and 909A9) placed Mdac within a 9.4-Mb interval region between D13Mit113 and D13Mit310. Both offspring exhibited a normal phenotype despite carrying the Dac mutant allele, indicating the presence of Mdac.

still absent despite a lack of dactylaplasia phenotype (Dac2J/ Dac mutant line showed consistent results, indicating that Dac1J Dac2J Mdac/mdac)[supporting information (SI) Fig. 5]. and Dac2J are both sensitive to Mdac. To investigate whether or not Mdac affects other loci, we looked for other unmethylated LTR retrotransposons in the Discussion mouse genome. A genome-wide in silico search for young We have characterized the Dac1J and Dac2J insertions from Dac elements, which have 100% identical 5Ј and 3Ј LTRs and display mutant alleles on mouse chromosome 19. These insertions were insertional polymorphisms in mouse strains, resulted in the identified as almost identical MusD elements, containing ORFs identification of another unmethylated LTR retrotransposon for gag, pro, and pol proteins of D-type virus. MusD is known to (ETnII) on mouse chromosome 17. The ETnII was present in the trigger the mobilization of ETn elements, which are the most Dac1J parental strain (NW࿝001030622) but was absent in the active murine mobile elements and cause the majority of inser- Dac2J parental strain and C57BL/6J (NT࿝039649). When com- tional mutations in mice (4, 5, 20, 21). Although the MusD itself pared with an unmethylated 5Ј LTR of the ETnII in the Dac1J has been shown previously to be autonomous for transposition parental strain (mdac/mdac), the 5Ј LTR was slightly methylated in a tissue culture assay, no pathegnic MusD insertion has been in F1 hybrids of the Dac1J strain and C57BL/6J (Mdac/mdac), identified to date (20). We have identified in vivo a MusD exhibiting two distinct populations of PCR clones (SI Fig. 6). element as a de novo and pathogenic insertion. It is noteworthy that two independent MusD insertions were Mapping of Mdac. Mdac has been mapped to a 28-Mb region in the identified at the same locus; however, it is still unclear how these middle of chromosome 13, between D13Mit10 and D13Mit99 (2). MusD insertions lead to the dactylaplasia phenotype. In the GENETICS To refine the Mdac locus, the dactylaplasia mouse (Dac/ϩ Dac2J mutant allele, the MusD element was inserted at intron 5 mdac/mdac) was crossed with C57BL/6J (ϩ/ϩ Mdac/Mdac)to of the dactylin gene. Owing to this mutation, dactylin transcripts produce F1 hybrids (Dac/ϩ Mdac/mdac). The F1 hybrids were are absent, suggesting that disruption of dactylin causes the then backcrossed to the dactylaplasia parental strain (Dac/ϩ dactylaplasia phenotype. On the other hand, the Dac1J insertion, mdac/mdac). These backcrosses were typed for both Dac and which resides in the upstream region of dactylin, affects neither microsatellite markers on chromosome 13, and phenotypes were the amount nor the size of the dactylin transcript (3). Further- compared. We analyzed 309 offspring obtained from the back- more, the dactylin transcript was also absent in Dac2J mice crossing test (207 of Dac1J line and 102 of Dac2J line). Eleven of carrying the Mdac allele (Dac2J/Dac2J Mdac/mdac), which show 309 offspring exhibited recombination events that were infor- no dactylaplasia phenotype (SI Fig. 5). These data suggest that mative for further mapping between D13Mit10 and D13Mit99. dactylin transcript levels are not essential for the dactylaplasia Among these, two independent recombination events placed phenotype. Mdac within a 9.4-Mb interval region between D13Mit113 and The dactylaplasia mouse is an ideal phenotypic and genotypic D13Mit310 (Fig. 4). Two mice (922A1 and 909A9) exhibited a model for human ectrodactyly (SHFM3) because the responsible normal phenotype despite carrying the Dac mutant allele. These SHFM3 locus on human 10q24 is homologous to the Dac locus mice led us to exclude markers that were homozygous for the on mouse chromosome 19 (2, 6–9). Recently, genomic rear- dactylaplasia parental strain, because both mice must have rangements have been identified in several unrelated SHFM3 inherited Mdac from the C57BL/6J strain. Backcrossing of each families (10–13). Each was a tandem genomic duplication con-

Kano et al. PNAS ͉ November 27, 2007 ͉ vol. 104 ͉ no. 48 ͉ 19037 Downloaded by guest on September 30, 2021 sisting of a disrupted extra copy of the dactylin gene and the not only in maize (31) but also in Arabidopsis (32, 33), Drosophila entire LBX1, BTRC, POLL, and DPCD genes. No other muta- (34), and mice (25, 35, 36). Although many other controlling tions have been identified in SHFM3 patients to date. Overdos- elements have been identified and characterized in different age of genes within the duplicated locus might be responsible for species, the mechanism by which the host genome regulates these SHFM3. Similarly, the Dac insertion might cause overexpression controlling elements is not well understood. of these genes. LTR retrotransposons are known to affect Our backcross study has narrowed the Mdac/mdac locus to a transcription of neighboring genes because the LTR has both 9.4-Mb interval between D13Mit113 and D13Mit310 on mouse sense and antisense promoter activity (22). We quantified LBX1, chromosome 13. Bisulfite sequencing and ChIP studies suggest BTRC, POLL, and DPCD mRNA levels in dactylaplasia embryos that Mdac is a DNA methylation, histone-modifying enzyme or but saw no significant changes in expression relative to WT (data a molecule involved in the RNAi pathway that recognizes not shown). retrotransposons; however, no known enzymes have been Aberrant expression of MusD elements in the AER was mapped in this region. The target of Mdac,orhowMdac observed in dactylaplasia mutant embryos (Dac/ϩ mdac/mdac or recognizes it, is unclear. Most LTR retrotransposons are meth- Dac/Dac mdac/mdac) but not in WT embryos (ϩ/ϩ mdac/mdac) ylated and suppressed in mice regardless of Mdac status, sug- or dactylaplasia embryos carrying Mdac (Dac/Dac Mdac/mdac). gesting that Mdac only recognizes active MusD and/or other These observations suggest the possibility that the Dac insertion LTR elements or that Mdac activity depends on genome position itself is expressed in the AER of mutant limb buds, although our and is effective only around the dactylin gene. probe detects not only transcripts from the Dac insertion but also During the study of another unmethylated ETnII element in other MusD and ETn transcripts because of sequence similarity. the Dac1J parental strain, the 5Ј LTR of the ETnII exhibited a The Dac insertion is presumed to be poorly transcribed in the slight increase in DNA methylation when crossed with C57BL/ presence of Mdac because of its heavily methylated status. More 6J. Mdac may affect other active LTR retrotransposons regard- interestingly, the ectopic expression seems to be correlated with less of their genomic locations; however, we cannot rule out the spatiotemporal loss of Fgf8 expression in the AER (19). De- possibility that the C57BL/6J strain carries another modifier creased expression of Fgf8 in the dactylaplasia mouse is thought which recognizes this ETnII. Transgenic mouse study of Mdac in to be caused by degeneration of the AER. Fgf8-null mice result the mdac/mdac background would be necessary to investigate in early embryonic lethality, whereas heterozygous Fgf8 mutants the effect of Mdac on other epigenetic sensitive loci. show normal limb development. On the other hand, conditional Of great interest are the observations that mouse dactylaplasia knockout studies demonstrated that Fgf8 deficiency in the AER and human SHFM3 are caused by a MusD insertion and genomic causes hypoplasia or aplasia of the limb (23, 24). duplication, respectively. Further study is necessary to under- It is noteworthy that Fgf8 resides on mouse chromosome 19, stand the link between dactylaplasia and SHFM3 and to elicit the only 70 kb away from the dactylin gene in head-to-tail orienta- mechanisms underlying these malformations. Moreover, identi- tion (Fig. 1A). An intracisternal A particle insertion in the fication of Mdac will shed light on the mechanisms by which host nearby agouti gene, which controls coat color, is known to cause silence transposable elements. aberrant expression of the agouti gene. Different methylation states of the intracisternal A particle insertion in different cells Materials and Methods lead to patchy coat color (25). Therefore, it cannot be ruled out All primer sequences are listed in SI Table 1. All animal studies that the active Dac insertion affects its neighboring gene, Fgf8. were performed in compliance with Osaka University guidelines. An external transcript from the Dac insertion toward Fgf8, which would be antisense for Fgf8, could cause Fgf8 repression in trans Isolation and Sequencing of Dac Insertions. We amplified each Dac by the RNAi machinery. We investigated this possibility by insertion by using long primers that flanked the insertion. PCR conducting RT-PCR and whole mount in situ hybridization; was carried out in a final volume of 50 ␮l containing 1ϫ LA PCR however, we failed to detect antisense Fgf8 transcript or any buffer II (Mg2ϩ free; Takara, Shiga, Japan), 200 nM each primer, transcription from the Dac insertion to Fgf8 (data not shown). 250 ␮M each dNTP, 500 ng of genomic DNA, 2.5 mM MgCl2, The 5Ј LTR of the Dac insertion showed two contrasting and 2.5 units of Ex TaqDNA polymerase (Takara) by using the epigenetic states that were dependent on the presence of Mdac. GeneAmp PCR system 9700 (Applied Biosystems, Foster City, Although most retrotransposons are heavily methylated and CA). Cycling conditions were as follows: initial denaturation at inactivated in somatic cells, little is known about the mechanism 95°C for 1 min, followed by the two-step profile: denaturation at by which the host genome suppresses transposable elements. 98°C for 10 s and annealing–extension at 68°C for 20 min for 10 Potential strategies include transcriptional silencing (DNA cycles, then autoextension by 20-s per cycle for an additional 20 methylation and/or chromatin modification), posttranscriptional cycles. PCR products were purified by using the QIAquick PCR silencing (RNAi), and mutational inactivation (cytosine deami- purification kit (Qiagen, Valencia, CA) and sequenced directly nases). Some apolipoprotein B mRNA editing complex by using BigDye Terminator v3.1 (Applied Biosystems) and an (APOBEC) proteins are known to suppress LTR retrotrans- internal primer. posons, including the intracisternal A particle and MusD, by either deaminase-dependent or unknown mechanisms (26–28). Bisulfite Sequencing. DNA (3 ␮g) extracted from embryonic tissue APOBEC proteins work primarily on transcribed LTR retro- (E9–E11.5) was digested with EcoRI and treated with sodium transposons by inducing mutations in the transposed copies or by bisulfite according to standard protocols. The bisulfite-treated reducing cDNA levels. Thus, Mdac might differ from the DNA was resuspended in 30 ␮l of TE [10 mM Tris/1 mM EDTA APOBEC family because Mdac appears to work as a pretran- (pH 8.0)], 2 ␮l of which was used in the PCR to amplify each scription barrier, causing DNA and/or histone methylation. LTR. PCR fragments were subcloned into the pCR4-TOPO The MusD insertion seems to act as a ‘‘controlling element’’ vector (Invitrogen, Carlsbad, CA) and then sequenced by using in the dactylin locus. Barbara McClintock first discovered trans- M13Forward or M13Reverse standard primer. posable elements in the 1940s during studies of maize and called them ‘‘controlling elements’’ because they had the ability to alter Genotyping of Dac and Physical Mapping of Mdac. Genotyping of the normal patterns of in a variety of ways (29). Dac allele was performed by using multiplex PCR analysis. DNA These alterations are dependent on the activity state of the was isolated from the amniotic membrane or the tail. Noon of (30). Recent studies have provided details the day on which a vaginal plug was detected was considered of the molecular events underlying these epigenetic phenomena E0.5. To refine the Mdac locus, the dactylaplasia mouse (Dac/ϩ

19038 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0705483104 Kano et al. Downloaded by guest on September 30, 2021 mdac/mdac) was crossed with C57BL/6J (ϩ/ϩ Mdac/Mdac)to containing T7 (sense strand) or SP6 (antisense strand) RNA produce F1 hybrids (Dac/ϩ Mdac/mdac). The F1 hybrids were polymerase promoters. In situ hybridizations were performed then backcrossed to the dactylaplasia parental strain (Dac/ϩ according to standard protocols. The sense strand probe was SEE COMMENTARY mdac/mdac). These backcrosses were typed for both Dac and used as a negative control. A minimum of four embryos were microsatellite markers on chromosome 13, and phenotypes were analyzed for each genotype, and at least two separate experi- compared. ments were conducted for each probe. Microsatellite markers used for mapping of Mdac were D13Mit10, D13Mit54, D13Mit20, D13Mit113, D13Mit209, ChIP Assay and Quantitative Real-Time PCR. The ChIP assay was D13Mit283, D13Mit310, D13Mit66, D13Mit281, D13Mit26, performed according to the protocol provided by the manufac- D13Mit24, and D13Mit99, which were informative for the back- turer (Upstate Biotechnology, Lake Placid, NY), with slight crossing test between dactylaplasia strains and C57BL/6J. One modification. We prepared Ϸ3 ϫ 107 cells from homogenized primer of each pair was labeled by 6-carboxyfluorescein at the fresh embryos (E10.5) and fixed them in 1% formaldehyde. 5Ј end. Fluorescent PCR products were subjected to electro- Antibodies used for ChIP were anti-acetylated histone H4, phoresis on the gel and/or analyzed by Automated Fluorescent anti-acetylated histone H3, anti-dimethyl histone H3-Lys-4, and DNA Sequencer by using GeneScan software (ABI 3100; Ap- anti-dimethyl histone H3-Lys-9 (Upstate Biotechnology). DNA plied Biosystems). We excluded the following markers from the recovered from immunoprecipitated complexes was subjected to Mdac locus by the backcrossing test. (i) Homozygous markers of quantitative real-time PCR with SYBR Green PCR Master Mix a mouse which exhibits normal phenotype despite carrying the by using an ABI PRISM 7900HT (Applied Biosystems) as Dac mutant allele. This mouse must have Mdac (Dac/ϩ Mdac/ described previously (37, 38). Primers were designed to cover mdac or Dac/Dac Mdac/mdac). (ii) Heterozygous markers of a each LTR and the Actb promoter region. The data were sum- mouse which exhibits dactylaplasia phenotype. The mdac must marized after normalizing either to the MusD element on be homozygous to exhibit dactylaplasia phenotype (Dac/ϩ mdac/ AL773522 (anti-acetylated histone H4, anti-acetylated histone mdac or Dac/Dac mdac/mdac). Primer sequences used to amplify H3, and anti-dimethyl histone H3-Lys-4) or to Actb (anti- microsatellite markers were derived from information available dimethyl histone H3-Lys-9); levels for both were set to 1.0. ChIP was performed by using at least five embryos for each genotype, at Mouse Genome Informatics (www.informatics.jax.org). and PCRs were performed at least twice for each sample. Whole Mount in Situ Hybridization. For whole mount in situ hy- We thank Drs. Kazuhiro Kobayashi, Nobuhiro Fujikake, and Yoshitaka bridization, embryos were collected and fixed in 4% parafor- Nagai for helpful comments and Dr. Jennifer Logan for editing the maldehyde in PBS. Digoxigenin-labeled probes were generated manuscript. This work was supported by the 21st Century Centers of by transcription (Roche, Indianapolis, IN) from amplification Excellence program of the Ministry of Education, Culture, Sports, products generated by RT-PCR from embryonic total RNA and Science and Technology of Japan.

1. Chai CK (1981) J Hered 72:234–237. 18. Loebel DA, Tsoi B, Wong N, O’Rourke MP, Tam PP (2004) Gene Expr Patterns 2. Johnson KR, Lane PW, Ward-Bailey P, Davisson MT (1995) Genomics 4:467–471. 29:457–464. 19. Crackower MA, Motoyama J, Tsui LC (1998) Dev Biol 201:78–89. 3. Sidow A, Bulotsky MS, Kerrebrock AW, Birren BW, Altshuler D, Jaenisch R, 20. Ribet D, Dewannieux M, Heidmann T (2004) Genome Res 14:2261–2267. Johnson KR, Lander ES (1999) Nat Genet 23:104–107. 21. Maksakova IA, Mager DL (2005) J Virol 79:13865–13874. 4. Ostertag EM, Kazazian HH, Jr (2001) Annu Rev Genet 35:501–538. 22. Druker R, Whitelaw E (2004) J Inherit Metab Dis 27:319–330. 5. Baust C, Baillie GJ, Mager DL (2002) Mamm Genome 13:423–428. 23. Moon AM, Capecchi MR (2000) Nat Genet 26:455–459. 6. Nunes ME, Schutt G, Kapur RP, Luthardt F, Kukolich M, Byers P, Evans JP 24. Lewandoski M, Sun X, Martin GR (2000) Nat Genet 26:460–463. (1995) Hum Mol Genet 4:2165–2170. 25. Morgan HD, Sutherland HG, Martin DI, Whitelaw E (1999) Nat Genet 7. Gurrieri F, Prinos P, Tackels D, Kilpatrick MW, Allanson J, Genuardi M, 23:314–318. Vuckov A, Nanni L, Sangiorgi E, Garofalo G, et al. (1996) Am J Med Genet 26. Esnault C, Heidmann O, Delebecque F, Dewannieux M, Ribet D, Hance AJ, 62:427–436. Heidmann T, Schwartz O (2005) Nature 433:430–433. 8. Raas-Rothschild A, Manouvrier S, Gonzales M, Farriaux JP, Lyonnet S, 27. Chen H, Lilley CE, Yu Q, Lee DV, Chou J, Narvaiza I, Landau NR, Weitzman Munnich A (1996) J Med Genet 33:996–1001. MD (2006) Curr Biol 16:480–485. 9. Ozen RS, Baysal BE, Devlin B, Farr JE, Gorry M, Ehrlich GD, Richard CW 28. Esnault C, Millet J, Schwartz O, Heidmann T (2006) Nucleic Acids Res (1999) Am J Hum Genet 64:1646–1654. 34:1522–1531. 10. de Mollerat XJ, Gurrieri F, Morgan CT, Sangiorgi E, Everman DB, Gaspari 29. Fincham JR, Sastry GR (1974) Annu Rev Genet 8:15–50. P, Amiel J, Bamshad MJ, Lyle R, Blouin JL, et al. (2003) Hum Mol Genet 30. Whitelaw E, Martin DI (2001) Nat Genet 27:361–365. 12:1959–1971. 31. Martienssen R, Baron A (1994) Genetics 136:1157–1170. 11. Kano H, Kurosawa K, Horii E, Ikegawa S, Yoshikawa H, Kurahashi H, Toda 32. Rangwala SH, Elumalai R, Vanier C, Ozkan H, Galbraith DW, Richards EJ T (2005) Hum Genet 118:477–483. (2006) PLoS Genet 2:270–281. GENETICS 12. Everman DB, Morgan CT, Lyle R, Laughridge ME, Bamshad MJ, Clarkson 33. Rangwala SH, Richards EJ (2007) Genetics 176:151–160. KB, Colby R, Gurrieri F, Innes AM, Roberson J, et al. (2006) Am J Med Genet 34. Gdula DA, Gerasimova TI, Corces VG (1996) Proc Natl Acad Sci USA A 140:1375–1383. 93:9378–9383. 13. Lyle R, Radhakrishna U, Blouin JL, Gagos S, Everman DB, Gehrig C, 35. Duhl DM, Vrieling H, Miller KA, Wolff GL, Barsh GS (1994) Nat Genet Delozier-Blanchet C, Solanki JV, Patel UC, Nath SK, et al. (2006) Am J Med 8:59–65. Genet A 140:1384–1395. 36. Rakyan VK, Chong S, Champ ME, Cuthbert PC, Morgan HD, Luu KV, 14. Ianakiev P, Kilpatrick MW, Dealy C, Kosher R, Korenberg JR, Chen XN, Whitelaw E (2003) Proc Natl Acad Sci USA 100:2538–2543. Tsipouras P (1999) Biochem Biophys Res Commun 261:64–70. 37. Su RC, Brown KE, Saaber S, Fisher AG, Merkenschlager M, Smale ST (2004) 15. Mager DL, Freeman JD (2000) J Virol 74:7221–7229. Nat Genet 36:502–506. 16. Yoder JA, Walsh CP, Bestor TH (1997) Trends Genet 13:335–340. 38. Hashimshony T, Zhang J, Keshet I, Bustin M, Cedar H (2003) Nat Genet 17. Jaenisch R, Bird A (2003) Nat Genet 33(Suppl):245–254. 34:187–192.

Kano et al. PNAS ͉ November 27, 2007 ͉ vol. 104 ͉ no. 48 ͉ 19039 Downloaded by guest on September 30, 2021